454 FRUITS REPLACED BY OFFSHOOTS. offshoots. Similar in its behaviour is another Composite, the alpine Adenostyles Cacalioe. It blossoms and fruits in sub-alpine forests even up to the tree hmit, but in high alpine regions, above 2200 metres in altitude, it never flowers, but forms offshoots, and in this way fills httle depressions on alpine slopes with its vigorous foliage. The terrestrial form of Polygonum amphibium occurs in a little bog close to my country house in the Gschnitzthal in the Tyrol at a height of 1200 metres. For twenty-eight years I have examined this bog every year without ever finding a ripe fruit upon these plants. But it propagates itself with rare luxuriance by means of offshoots and forms a broad girdle around the bog. These plants, Nardosmia frigida, Adenostyles Cacalioe, and Polygonum amphibium,, grown in a more favourable climate, produce good seed, but their vegetative methods of propagation are so restricted that one might almost suppose them to be different species of plants. Instances in which flowers are replaced by offshoots or bulbils in the inflores¬ cence may be mentioned in connection with the above. Polygonum vivi/parum and bulbiferum, Saxifraga cernua, nivalis, and stellaris, Juncus alpinus and supinus, and the Grasses Aira alpina, Festuca alpina and ru/picaprina, Poa alpina and cenisia occur, it is true, with normally developed flowers and fruits, but in alpine, and especially in arctic regions, where these plants have their head¬ quarters, one very frequently finds purely vegetative buds or bulbils, which become detached from the parent plant and give rise to new individuals, in place of flowers and fruit. In the Polygonums mentioned little bulbils replace a portion of the flowers. Saxifraga cernua usually produces a single terminal flower at the end of its inflorescence, the lateral flowers being replaced by little tufts of bud-like offshoots on short stalks (see fig. 342 ^). These buds, when they fall off, are either still closed (fig. 342 ^), or their thick, fleshy, outer scales are already parted, exposing a little green foliage-leaf. On the ground they soon produce roots and grow into new plants (see figs. 342 ^ and 342''). In Saxifraga nivalis little shoots are formed in place of flowers, each bearing a tuft-like rosette of minute leaves (flg. 342 '). These rosettes are readily separable, and producing roots from their abbreviated axes, give rise to new plants. So also in the Juncuses and Grasses mentioned, httle shoots replace the fruits and come awajr from the inflorescence. These shoots are produced in Poa alpina (see fig. 342 ^) and in most of the other Grasses mentioned, in the following manner. The axis of each spikelet, after producing several glumes at its base, forms green leaves above—as it were a grass-plant in miniature (see figs. 342^ and 342^"). Later, these disarticulate, take root, and grow into new plants. More rarely do shoots arise laterally on the axis, in the axUs of subtending scales; when this is the case they faU away in the usual manner. The earlier Botanists termed aU such Grasses, and indeed all plants which produce bulbUs in their inflor¬ escences, viviparous, the idea being, that in all of them the seeds germinated precociously whilst stiU attached to the parent. This view was probably suggested by the common experience of agriculturalists that Rye, Oats, and other cereals sometimes " sprout", i.e. that when the spikes are continually wetted by rain about